A Synoptic Weather Typing Approach To Simulate Daily Rainfall In Guangzhou Region

Urban waterlogging disasters caused by the heavy rainfall damages city system severely. Situated in the east Asian monsoon region, Guangzhou-centered city cluster is the strong rainfall and frequent rainstorm area. Conclusions conducted from this research has great significance in preventing urban waterlogging disasters in this region.In this thesis an automated synoptic weather typing and stepwise cumulative logit/nonlinear regression analysis are employed to simulate the occurrence and quantity of daily rainfall events for Guangzhou region. These daily rainfall simulation models have the potential to be used to predict changes in frequency and magnitude of future daily rainfall events. The synoptic weather typing is developed using principal component analysis, an average linkage clustering procedure, and discriminant function analysis to identify the weather types most likely to be associated with daily rainfall events.11 synoptic weather types are identified over the 23-yr period as the primary rainfall-related weather types among all the 57 synoptic weather types. Meteorological weather patterns associated with the 11 rainfall-related weather types are identified with the surface weather patterns, consisting of 5 weather patterns which situated in different position or occurring in different seasons (remarks Ⅰ,Ⅱ, and Ⅲ):quasi-stationary front, cold front, heat low, tropical cyclone, cold high. Days and precipitation intensity in different rainfall-related weather type have great difference. The heavy rainfall events are mainly distributed in the ahead of quasi-stationary front Ⅰ、Ⅱ, tropical cyclones and ahead of cold front whether types.Within-weather-type daily rainfall simulation models comprise a two step process:(ⅰ) cumulative logit regression to predict the occurrence of daily rainfall events, and (ⅱ) using probability of the logit regression, a nonlinear regression procedure to simulate daily rainfall quantities. The rainfall simulation models are validated using an independent dataset, and the results show that the models are successful at replicating the occurrence and quantity of daily rainfall events. For example, the relative operating characteristics score is greater than 0.98 for rainfall events with daily rainfall≥25 or≥50 mm for both model development and validation. For evaluation of daily rainfall quantity simulation models, four correctness classifications of excellent, good, fair, and poor are defined, based on the difference between daily rainfall observations and model simulations. The percentage of excellent and good simulations for model development ranges from 55% to 73% when simulating no/light rain, moderate rain, heavy rain and rainstorm, the corresponding percentage for model validation ranges from 40% to 72%. However, the percentage of excellent and good simulations is only 31% and 14% for model development and validation when simulating downpour. To determine whether it is necessary to perform synoptic weather typing prior to the applicationof a regression model, the rainfall simulation model is redeveloped using all days without synoptic weather typing. The results imply that the synoptic weather typing prior to the application of cumulative logit-nonlinear regressions improves the percentage of excellent and good simulations from 2% to 6% when simulating moderate rain, heavy rain and rainstorm.